Abstract:The formation flying of aviation swarm is a necessary condition for the realization of aviation cluster capability emergence. As for an aviation cluster formation with more aircraft, while the traditional leader wingman following mode being adopted, and as a result of a large number of wingmen, there is a chance that some wingman has a position deviation, aircraft-to-aircraft collision with the surrounding wingman may happen easily, and there is a great security risk. For this reason, a space alignment control method for large formation of aviation cluster is proposed, and the organizational structure of aviation cluster formation is constructed. Through the hierarchical alignment method and the reference aircraft selection algorithm based on distance, the overall formation of aviation cluster formation is achieved.

Abstract:In view of the influence of the front and rear edges of the wing on the static aeroelasticity of the canard-forward-swept wing configuration aircraft, through threedimensional unsteady N-S equations and linear elastic statics equations solved by the CFD/CSD loosely coupled calculation method, the aerodynamic and elastic characteristics of the elastic forwardswept wing are obtained under condition of the state of independent deflection and coordinated deflection of the leading and rear edges control surface. The results show that the elastic wing is characterized by better lift characteristics and high angle of attack stall characteristics compared with the rigid wing, and the changes of the control surface deflection method also affect the aerodynamic and elastic characteristics.When the control surface is deflected alone, both of the downward deflection of the leading edge control surface and the downward deflection of the trailing edge control surface can increase the lift coefficient of the elastic wing, and the maximum lift coefficient increments are 2.60% and 8.69% respectively. When the control surfaces are deflected in coordination, the lift coefficient in deflecting in the same direction is greater than that of single deflection, and the maximum lift increment is 11.96%. The liftdrag ratio of reverse deflection is better, and the elastic deformation and torsion can be reduced in the range of small angle of attack.

Abstract:A threedimensional finite element model of the joint assembly is built, and the failure evolution process of the composite laminate is analyzed. The numerical analysis model is modified by regulating initial stiffness coefficients of the laminate based on the strength test data. The joint strength of the modified numerical model is fitted well with that of static loading test. It follows that the approximate dual linear characteristics is found in the tensile test curves before the ultimate strength is met, and tensile load attenuation surge happens at the point where the gradient discontinuity occurs. The contact faces of the tensile test specimen encounter a process from initial static friction state, transient slide friction state to final static friction equilibrium state. When there is a M6 bolt with 8 N·m tightening torque, its tensile strength is the maximum, and the relative errors of joint strength between the numerical analysis results and that of the tensile test are all less than 6.7%. The best joint strength can be obtained to proper tightening torque, whereas excessive tightening torque may lead to strength reduction.

Abstract:Aimed at the problems that aero-engine is complex in structure, severe nonlinearity of various degenerate state is variable, and traditional physical failure model-based method is difficult to predict the remaining useful life of the engine (Remaining Useful Life, RUL) accurately, the problems abovementioned can be done by adopting an improved convolution neural networks (CNN). A linear degradation model is employed to label each sample. The convolution is set to several different onedimensional convolutions to extract data features and the correlation between the RUL better. In order to validate the effectiveness of the method, a test is made on the commercial modular aeropropulsion system simulation (C-MAPSS) aircraft engine datasets provided by NASA. The results show that the convolutional neural network has higher precision compared with the common neural network.

Abstract:With low infrared emissivity coating (LIREC) being easy to get bubble, shedding, wear out, scratch, dirt and other problems, the performance state of LIREC should be measured and evaluated in time to ensure the long service life of the LIREC. A portable infrared emissivity measuring instrument and a portable infrared thermal imager used to test the emissivity of shedding, scratching and fouling coating areas at 5 kinds of temperature are adopted. The result shows that the influence of shedding or fouling on the LIREC is comparatively larger, the scratching and the fouling also make the emissivity of LIREC larger. Therefore, the influence degree should be timely and quantitatively measured and evaluated. The measurement results tested by portable infrared emissivity tester are affected by the surface conditions of the measurement area such as surface roughness and so on. The measurement results tested by portable infrared thermal imager are more affected by temperature.

Abstract:Aimed at the problems that existing software defined network (SDN) multi-controller domain partitioning algorithms are mostly based on the ground networks with fixed topology, difficult to adapt to the satellite network with dynamic topology, and most of the existing satellite network researches are only in consideration of the propagation delay, a software defined satellite network (SDSN) delay analysis model is proposed in view of processing delay and queuing delay. On the basis of this model, satellite network multicontrol domains are divided. The simulations prove that this algorithm can adapt to the satellite network environment better compared with the existing ground SDN multicontrol domain planning algorithm and the satellite network control domain division method. Besides, the algorithm can effectively reduce the network delay, and increase the load balancing between the controllers by more than 40%. Moreover, the iteration stopping mechanism makes the algorithm run fast, and can adapt to the rapid changes of the satellite network topology.

Abstract:The existing human target recognition technology based on the traditional planar electromagnetic wave radar is enabled to realize good results in the face of the classification and recognization of human targets with large gait difference, only a great difficulty remains in the fine recognition of human gait. In this paper, the singlefrequency vortex electromagnetic wave radar (VEMWR) is applied to human gait recognition, and the target information in radar echo is increased by transmitting the vortex electromagnetic wave with orbital angular momentum (OAM), thus improving the ability of human gait fine recognition. A singlefrequency VEMWR echo model of human target is established, and the echo data sets of three kinds of gait are generated by the simulation. Through transforming the echo into the fundamental frequency, the linear Doppler and angular Doppler mixed information are obtained, and then their timefrequency analysis images being input into a convolution neural network model, the classification results are obtained. The simulation results show that compared with the traditional planar electromagnetic wave radar, the use of vortex electromagnetic wave is effective in improving the fine recognition ability of human gait.

Abstract:Aimed at the characteristic that the error rate of NAND FLASH increases with the increase of the time of use, a highspeed parallel RCRF+BCH error correction scheme with stronger error correction capability while using fewer parity bits is proposed. The idea of RCRF corrects some of the initial erasure errors, and then cascades BCH codes to correct the remaining bit errors, which can greatly ensure the accuracy of data and significantly improve the reliability of the storage system. The article explains in detail the encoding and decoding principles and execution steps of the highspeed parallel algorithm. In the BCH part, the iBM key equation solving algorithm without inversion that consumes less hardware resources is used, and then the different paths of the error position polynomial are listed through derivation Several-deterministic forms facilitate the application of combinatorial logic to describe them, thus avoiding the complicated iterative judgment process and further improving the decoding speed. And adopt the modular processing method and the pipeline operation mode to optimize the BCH codec structure. Finally, it was implemented on FPGA platform hardware and simulated to verify the effectiveness of this scheme.

Abstract:In view of fusion object detection of visible light image and infrared image, an object detection algorithm based on decisionlevel fusion is proposed. The YOLOv3 network is retrained by establishing a labeled data set, and the trained YOLOv3 network is used to detect the visible light image and the infrared image separately before fusion. In the process of fusion, a novel detection fusion algorithm is proposed. First, the accurate results of the object detected only in the visible image or only the infrared image are retained, and then the accurate results of the same object in the object detected in both of the visible image and the infrared image are weighted and fused. Finally, taking the combined detection results as the detection results of all corresponding objects in the fused image, the rapid object detection based on decisionlevel fusion is realized. The experimental results show that all indicators have a good performance on the established data set. Among them, the detection accuracy of the proposed algorithm amounts to 84.07 per cent. Compared with the algorithm of detecting the visible light images and the infrared images alone, the detection accuracy of the algorithm increases by 2.44% and 21.89% respectively, more objects can be detected, and there is a decrease in false detection of objects. Compared with the three detection algorithms based on featurelevel image fusion, the detection accuracy of the algorithm increases by 4.5%, 1.74 %, and 3.42% respectively.

Abstract:Millimeter wave is high in path loss, strong in directivity, and is very sensitive to the obstacles caused by obstacles and the Doppler frequency shift caused by motion. Therefore, how to provide a stable link quality becomes a problem to be solved in millimeter wave communication systems. Aimed at the problems that electromagnetic waves have different propagation characteristics in consideration of the continuous increase in the number of antennas and cost, traditional multiuser precoding technology is difficult to use in millimeter wave communication systems, and the combination of millimeter wave communications with the Massive MIMO has become an important research direction for future wireless communication development, a joint algorithm of user frequency domain scheduling in the millimeter wave system is proposed based on FDMA and mixed precoding of multiple users, improving the performance in the millimeter wave communication system. The simulation results show that this algorithm is similar to the pure digital precoding in performance. Compared with other similar algorithms, the increase in computational complexity is not obvious, and the algorithm can greatly improve the performance in the millimeter wave Massive MUMIMO system.

Abstract:UAV collaborative operation is characterized by the advantages of high mission completion and low operational consumption, and is one of the main trends in future operations. Effectiveness assessment of UAV cooperative operations can be effectively used to analyze the performance of the system and has a certain of positive significance for improving the combat capability of the system. In view of making the formation of unmanned cooperative combat system by using DoDAF (Department of Defense Architectural Framework), the relationships between the combat activities and the capabilities are analyzed, a coordinated combat capability index system is established. different types of indexes Processing are differentiated respectively by using scale scaling method, probability calculation method, and aggregation method to deal with dimensional indicators, quantitative probability indicators and qualitative probability indicators. ADC (availability dependability capability) model is utilized for expanding system capability vectors, taking into account the comprehensive indicators, such as survivability and electronic confrontation capabilities, breaking through the limitations of single indicators. Finally, taking the UAV collaboration under the typical mission of ground combat as a scenario, by the method proposed in this paper, the effectiveness of different combat options is evaluated and calculated. The calculation results verify the effectiveness of the method. And the method is valid.

Abstract:In radar target adaptive detection problems, parts of the samples in the reference data are missing, such case is as a result of the degradation of detection performance for conventional detectors. Being under no condition of a prior information, an expectation maximization based adaptive matched filter (EM-AMF) is obtained by the maximal likelihood estimate of the clutter covariance matrix. A prior information about the detection environment is utilized for obtaining Markov Chain Monte Carlo based adaptive matched filter (MCMC-AMF) under condition of using Gibbs sampler to obtain the posterior mean of the clutter covariance matrix. The computer simulation results show that both of the detectors perform well under condition of the missing samples. With less prior information, MCMC-AMF and EM-AMF possess the analogous detection performance. But the MCMC-AMF detection performance can be improved further by exploiting more prior information.

Abstract:Aimed at the problems that being caused by deleting the repeated virtual array during DOA estimation, effective information and estimation performance are lost and poor, a symmetric nested MIMO radar DOA estimation algorithm is proposed based on redundancy average of virtual array. Firstly, a set of uniform linear array and a set of sparse uniform linear array are arranged symmetrically by taking zero as a center to constitute a monostatic MultipleInput MultipleOutput radar transmitting array and a receiving array. The array structure is to change the conventional virtual array for a symmetric sum coarray structure from the difference coarray structure, thus increasing the degree of freedom, reducing the mutual coupling of the system, and applying it in uncorrelated and coherent targets DOA estimation. Secondly, the vectorized sample covariance matrix and the Toeplitz matrix are reconstructed by averaging the redundant virtual elements of sumdifference coarray. Finally, the combination of the proposed algorithm with the MUSIC algorithm for DOA estimation effectively improves the performance of target estimation number and angle estimation. The simulation results show that the proposed array structure and algorithm are effective.

Abstract:1DT is a typical post Doppler space-time adaptive processing method, and the method can significantly reduce the demand for independent identically distributing to the samples and the amount of calculations for spacetime adaptive processing. However, being poor in the process of the mainlobe clutter suppression in the temporary domain, a filter design algorithm with weighted sparse constraints by using the sparse characteristics of the filter frequency response is proposed. A sparse constraint on the sidelobe is added to the model of the time domain filter, and a weighting matrix is constructed to shape the frequency response of the mainlobe clutter. The simulation experiments show that this method can reduce the sidelobe response under condition of effectively suppressing the mainlobe clutter presupposition, and is robust with the presence of array amplitudephase error.

Abstract:Elastic support is a commonly used means to the vibration isolation and the vibration reduction. Research on the dynamic response of elastically supported underground arch structures under under condition of explosive loads is of great significance for improving the battlefield survivability of personnel and equipment in underground protection projects. In this Paper, the ANSYS/LSDYNA nonlinear explicit dynamic finite element analysis program is applied to simulate the modal and dynamic response of underground arch structure with elastic support under condition of blast load, thus obtaining the low order natural frequency variation curve and the dynamic response time history curve of arch crown, arch shoulder and arch foot. The results show that the elastic support reduces the natural frequency of the structure, extends the natural vibration period of the structure, and has a good vibration isolation and vibration reduction effect. With the critical stiffness coefficient leading the first and the second modes of the structure to be easily excited at the same time, such a case should be avoided in the structural design. Compared with rigidly supported underground arches, the elastically supported underground arche is low in pressure and stress peaks, long in their appearance, good in vertical displacement peaks, and improvement in blast resistance. As for the stiffness coefficient of the elastic support, this does not mean the less the better without principle, it should be done reasonably according to the structural antiblasting capacity and ultimate displacement design requirements.

Abstract:Aimed at the problems that PM2.5 concentration prediction model is unstable in efficiency and poor in generalization ability, a TwoDirection Attentionbased Recurrent Neural Network (TDA-RNN) is proposed based on the cyclic neural network and attention mechanism. Firstly, the temporal attention and the category attention in inputting data through attention mechanism are obtained in making them fused by TDA-RNN model, and then the fused data are encoded to obtain intermediate features through feature encoder. Finally, the intermediate features are fused with historical information of PM2.5 concentration, and the predicted values are obtained by feature decoder. The PM2.5 concentration in Beijing is predicted by several models. The results show that the prediction accuracy of TDA-RNN is higher than that of the Back Propagation Neural Network, the Long ShortTerm Memory, the Gate Recurrent Unit and the Moving Average model. In the antijamming test, while the input data having noise factors, the prediction accuracy of TDA-RNN decreases slightly, but still higher than that of other models. The TDA-RNN proposed in the paper is strong in feature extraction ability and high in prediction accuracy.And this can also be applied to multivariate time series prediction in other application scenarios.